Motion determination for volumetric magnetic resonance imaging using a deep machine-learning model

A method for determining motion artifact for a magnetic resonance system, the method comprising:

• reconstructing a three-dimensional representation of a patient from a magnetic resonance scan of the patient;
• determining a level of the motion artifact from a machine-learned network in response to input of the three-dimensional representation to the machine-learned network;
• and
• displaying the level of the motion artifact with an image from the three-dimensional representation.

A method for machine learning to determine motion artifact, the method comprising:

• creating training data from a first magnetic resonance reconstruction from k-space data generated using a scan acquisition order applied to a volumetric magnetic resonance reconstruction subject to motion from a motion model;
• determining an error of the first magnetic resonance reconstruction to the volumetric-slice magnetic resonance reconstruction;
• training, by a machine, a neural network to receive a second magnetic resonance reconstruction from a magnetic resonance scan and to output a score, the training using the error and the training data;
• and
• storing the machine-learned neural network.